{"id":"W4401609584","doi":"10.1109/icasspw62465.2024.10626010","title":"Characterizing the Temporal Dynamics of Universal Speech Representations for Generalizable Deepfake Detection","year":2024,"lang":"en","type":"article","venue":"","topic":"Speech Recognition and Synthesis","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"ca_institutions":"Institut National de la Recherche Scientifique","funders":"","keywords":"Dynamics (music); Computer science; Speech recognition; Artificial intelligence; Natural language processing; Acoustics; Physics","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002216115,0.00005675643,0.00006630891,0.00009558626,0.0001001302,0.0001403358,0.0002253426,0.00003128349,0.00005755911],"category_scores_gemma":[0.00002688964,0.0000410606,0.00008116628,0.0003121354,0.0000232203,0.0003345912,0.00004508071,0.00004311154,0.00001407389],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004108225,"about_ca_system_score_gemma":0.00003624129,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001389989,"about_ca_topic_score_gemma":0.0003641488,"domain_scores_codex":[0.9994425,0.0000344109,0.0001373087,0.0001766081,0.0001060639,0.0001030687],"domain_scores_gemma":[0.9995236,0.000148409,0.00003430099,0.0002031032,0.00006689673,0.00002363342],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00001241883,0.00004671793,0.0001629323,0.00006339787,0.00007916945,0.0000115167,0.0006706558,0.00003574823,0.02828513,0.1543307,0.001230935,0.8150707],"study_design_scores_gemma":[0.00008258065,0.00002706491,0.0001764139,0.00001339579,0.00001243171,0.00002471761,0.0002513944,0.8992134,0.08994376,0.003978827,0.006203071,0.00007299917],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01452886,0.00001835035,0.978511,0.002274968,0.000459553,0.0001784632,0.00001051096,0.0001566033,0.003861684],"genre_scores_gemma":[0.8158793,0.00002299845,0.1788268,0.0002120146,0.0001100337,0.00003147765,0.00002568392,0.0000130528,0.004878653],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8991776,"threshold_uncertainty_score":0.1674403,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0281430471174211,"score_gpt":0.2683847684010593,"score_spread":0.2402417212836382,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}